OpenAI made Automations generally available in the Codex app with per-run model selection, reasoning controls, worktree or branch targeting, reusable templates, themes, and terminal visibility. Use it for unattended repo maintenance instead of limiting Codex to one-off interactive tasks.
git worktree prune, and proposed a weekly Friday cleanup job.OpenAI’s Codex update says Automations are now generally available in the app, and the new controls are the technically important part: users can choose the model and reasoning level for each run, decide whether execution happens in “a worktree or an existing branch,” and save workflows as templates for reuse. That turns Automations from a simple scheduler into a configurable execution layer for recurring repo tasks.
The rollout details in the rollout thread add the intended use cases: “auto-triage issues, PRs, and daily repo briefings.” The same thread says each automation can run unattended after you “define the task once, set a schedule,” which is the clearest signal that OpenAI wants Codex handling background maintenance and reporting work rather than only interactive coding sessions.
The most concrete implementation detail comes from the rollout thread, which says “each automation spins up a dedicated Git worktree in the background.” That matters operationally because it isolates agent work from the main branch while still letting the app act on a real repository state.
A practitioner demo in worktree cleanup demo shows the pattern on a live repo. Codex reported it had “removed 49 detached vault worktrees,” ran git worktree prune, and verified that “dirty changes in the main vault repo were not touched.” In the same flow, the user asked for a recurring cleanup job and Codex drafted a “Weekly worktree cleanup” automation scheduled for Fridays at 9:00 AM, with the repo path and cleanup scope spelled out in the generated description. That is a stronger proof point than the launch copy: the app is not just scheduling prompts, but composing repo-specific maintenance jobs around Git worktrees.
Codex can now “read the integrated terminal,” according to the terminal post. The example shown there has the user ask, “What’s in the terminal?” and Codex replies with the current shell, working directory, branch, and the last visible command output, including the cowsay "Hello Codex" result. That closes a common context gap for desktop coding agents: the model can inspect terminal state without the user pasting it into chat.
The other visible change is theming. OpenAI’s Codex update says users can import themes they like, share their own, and personalize fonts, colors, and contrast; screenshots in the Claude theme and the Sublime theme show community-made presets already circulating. OpenAI is also actively soliciting shared themes in the theme contest, which suggests the theme system is not a hidden preference panel but a supported part of the app’s workflow surface. A smaller but practical workflow tweak appears in the Raycast tip, where a user maps a Raycast deeplink to open Codex with selected text, cutting down on copy-paste when turning an error message or plan into a new task.
Claude can now drive macOS apps, browser tabs, the keyboard, and the mouse from Claude Cowork and Claude Code, with permission prompts when it needs direct screen access. That makes legacy desktop workflows automatable, and Anthropic is pairing the push with more background-task support for longer agent loops.
releaseOpenClaw shipped version 2026.3.22 with ClawHub, OpenShell plus SSH sandboxes, side-question flows, and more search and model options, then followed with a 2026.3.23 patch. Teams get a broader plugin surface, but should patch quickly and review plugin trust boundaries as the ecosystem grows.
releaseCursor shipped Instant Grep, a local regex index built from n-grams, inverted indexes, and Bloom filters that drops large-repo searches from seconds to milliseconds. Faster candidate retrieval shortens the coding-agent loop, especially when ripgrep-style scans become the bottleneck.
breakingChatGPT now saves uploaded and generated files into an account-level Library that can be reused across conversations from the web sidebar or recent-files picker. It removes repetitive re-uploading and makes past PDFs, spreadsheets, and images part of a persistent working context.
breakingEpoch AI says GPT-5.4 Pro elicited a publishable solution to one 2019 conjecture in its FrontierMath Open Problems set, with a formal writeup planned. Treat it as an early milestone worth reproducing, not blanket evidence that frontier models can already automate math research.
Sometimes we ship features so fast we forget to share them 😅 Codex in the app can now read the integrated terminal! Thanks @ajambrosino!
I’m constantly copying command output from the app command window. Shouldn’t codex be aware of the command output?
OpenAI announced two updates to the Codex app. 🔹Users can now personalize the app with custom themes, including importing themes they like or sharing their own. 🔹Automations are now generally available, with new controls that let users choose the model and reasoning level, Show more
We’ve been cooking. 2 updates in the Codex app 👇 You can now personalize the Codex app with themes that match your taste. Import themes you like or share your own.
make codex app your own
We’ve been cooking. 2 updates in the Codex app 👇 You can now personalize the Codex app with themes that match your taste. Import themes you like or share your own.
